I’m familiar with the game-theoretic argument against FPTP you’re citing to rebut Katy Levinson’s call to vote third party, and agree with you ten thousand percent that making structural changes should be a major, mainstream issue that we should all be talking about. But as a nitpick, I think your title is phrased a little too strongly, and as a result doesn’t really function as much of a rebuttal at all.

Basically, the issue is this. Yes, it seems likely that electoral systems based on plurality voting should tend to resolve into two-party systems… for the most part and over a long enough timescale. But there are plenty of counterexamples and exceptions, including the Philippines (primarily FPTP, multiple parties since 1987) and Canada (FPTP, currently three parties, which is less than normal). Duverger’s law isn’t really a law at all, nor was it ever intended to be understood as one; it’s a more general observation that plurality voting tends to exert an inhibiting effect on the growth of minor parties overall, although the magnitude of that effect varies based on a range of contextual factors.

In my view, the salient example with reference to Katy’s piece actually comes from US history. Under the First Party System (1792–1824), we had what appeared to be a two-party duopoly: the Federalists and the Democratic-Republicans were the only legit contenders — but they eventually fractured, if at different times and in different ways. Under much of the Second Party System (1828–1854), the Democratic Party and the Whigs dominated. The election of 1854 saw another big reshuffle, with a few parties forming and collapsing before some of those fragments coalesced into a coalition that would become the modern Republican Party.

Moral of the story? Sure, per Duverger FPTP seems to contain an inherent structural bias against minor parties, which can be expected to result in a two-party system over a long enough timeline. But it sure as shit doesn’t have to be these two parties. For well over half a century — about the entire first third of US history, if you start counting from the signing of the Constitution in 1787 — there were either no formal political parties, or regularly-changing major parties.

Should we get rid of FPTP? Abso-fucking-lutely. But in no way does FPTP ensure the continued dominance of the Democrats and the GOP — nor does it preclude electoral successes for third parties, up to and including a third party displacing the Democrats or Republicans and becoming a major party itself.

What FPTP does do, however, is give the major parties a significant structural advantage —and it’s hard to see why they would willingly relinquish that advantage in order to change the voting system. So getting back to Katy’s piece, there’s a fairly strong argument to be made that’s basically the reverse of the one you present: namely, that the best way to get rid of FPTP may well be to try to drive support for third parties willing to ditch it, rather than the other way around.